Proc. Natd Acad. Sci. USA Vol. 79, pp. 193-197, January 1982 Neurobiology

Benzodiazepine receptor increases after repeated seizures: Evidence for localization to dentate granule cells (kindling/epilepsy/drug receptor//) FERNANDO VALDES*t, RICHARD M. DASHEIFFt, FAITH BIRMINGHAMt, KEITH A. CRUTCHER§, AND JAMES 0. MCNAMARA*t¶II *Epilepsy Center, Veterans Administration Medical Center, Durham, North Carolina 27705; Departments of tMedicine (Neurology) and fPharmacology, Duke University Medical Center, Durham, North Carolina 27710; tSchool of Medicine, Duke University Medical Center, Durham, North Carolina 27710; and §Department of Anatomy, School of Medicine, University of Utah, Salt Lake City, Utah 84223 Communicated by James B. Wyngaarden, August 27, 1981

ABSTRACT Repeated seizures, whether induced by kindling alograms were recorded before and after each stimulation with or electroshock, result in increased numbers of benzodiazepine a Grass model 78 polygraph. Kindling developed as described receptors in hippocampal formation membranes. We sought to (6). Experimental animals experienced 18-23 stimulation-in- determine the cellular constituents containing the receptor in- duced seizures. Each animal experienced at least one class 5 creases. Binding studies of microdissected samples localized, the seizure consisting of rearing and falling. The number of stim- receptor increases to fascia dentata. [3H]Flunitrazepam autora- ulations required to elicit the first class 5 seizure was 13.1 ± diographic studies showed increases of silver grain density over 1.4 (mean ± SEM). Control animals underwent electrode im- the granule cell and molecular layers of fascia dentata but not in plantation but were not stimulated. Electrode placements were other regions of hippocampal formation. Destruction of granule not histologically verified cells by colchicine or neonatal x-irradiation was associated with because the stimulated was marked decrease of benzodiazepine receptor binding. Together, used for other experiments. these results provide evidence for localization of the receptor in- Electroshock Seizures. These seizures were induced by ad- creases to the somata and dendritic tree of the granule cells. We ministering 110 V (60 Hz) for 1 sec through earclips at daily in- suggest that this cellular localization may provide a clue to the tervals, 5 days per week. These seizures usually consisted of network ofaltered neural circuitry underlying amygdala kindling. tonic extension ofall four limbs followed by clonic contractions and subsequent postictal unresponsiveness. Control animals The benzodiazepines are effective anticonvulsants in both an- were connected to earclips but received no current. imals and man (1, 2). Recent investigations have identified bind- Dissection and Membrane Preparation. Paired control and ing sites in brain membranes that likely represent the receptors experimental animals were killed by decapitation 24 hr after the mediating the pharmacologic action ofthese drugs (3). We pre- last experimental manipulation. After removal by-blunt dissec- viously showed that repeated kindled ** seizures result in in- tion, the hippocampal formations were sectioned at 600-gm in- creased numbers ofbenzodiazepine receptors bilaterally in the tervals with a McIlwain tissue chopper in a plane perpendicular hippocampal formation but not in neocortex (6). to the longitudinal axis. By using a dissecting microscope and The basic mechanisms underlying the kindling phenomenon a scalpel, the sections were carefully separated into fascia den- are obscure (7). The responsible neuronal modifications appear tata and hippocampal gyrus. The dentate hilus was included in to involve a selected network ofneural circuits in multiple brain the hippocampal gyrus sample. regions (8, 9). Identification of the precise spatial distribution The microdissected samples were pooled separately and ho- of the altered neural circuits is essential to understanding the mogenized in 1 ml of cold 0.32 M sucrose/i mM Tris citrate underlying cellular and molecular mechanisms. buffer, pH 7.1, in a Teflon and glass homogenizer. The super- The heterogeneous regional distribution of the benzodiaze- natant was removed by centrifugation at 43,500 X g for 20 min, pine receptor increases raised the possibility of a discrete cel- and the pellet was suspended in 2 ml ofcold 1 mM Tris citrate lular localization within hippocampal formation. We present buffer, pH 7.1, with a Polytron (setting 6 for 30 sec) and then evidence supporting localization ofthe benzodiazepine receptor stored at -80TC. The samples were thawed on the day ofassay. increases to the dentate granule cells. This may provide a clue Aliqouts were removed for protein determination according to to the network ofaltered neural circuitry underlying amygdala the method of Lowry et al. (10). kindling. Benzodiazepine Receptor Binding. Binding was measured with the radioligand [3H]flunitrazepam (specific activity, 86.4 METHODS Ci per mmol; 1 Ci = 3. x 101i becquerels; New England Kindling. Adult male Sprague-Dawley rats (Charles River Nuclear). Specific binding was defined as the difference in total Breeding Laboratories) weighing at least 200 g were used. A Abbreviation: Bmn, maximal binding value. bipolar electrode was stereotaxically implanted in the basola- IITo whom reprint requests should be addressed: Director, Epilepsy teral nucleus of the right amygdala under pentobarbital anes- Center, 508 Fulton Street, Veterans Administration Medical Cen- thesia as described (6). After a postoperative recovery period ter, Durham, NC 27705. ofat least 7 days, stimulations from a Grass S88 stimulator (400- ** Kindling is an animal model of epilepsy induced by electrical stim- to 900-PuA biphasic square-wave pulses,. 1-msec duration, 60 Hz ulation of the brain (4, 5). The term kindling refers to the phenom- for 1 sec) were administered at enon whereby repeated periodic initially subconvulsive electrical daily intervals. Electroenceph- stimuli induce progressive intensification of stimulus-induced sei- zure activity, culminating in a generalized clonic motor seizure. The publication costs ofthis article were defrayed in part by page charge Once established, this enhanced sensitivity to electrical stimulation payment. This article must therefore be hereby marked "advertise- is permanent. The amygdala is a sensitive region in that relatively ment" in accordance with 18 U. S. C. §1734 solely to indicate this fact. few stimulation episodes are required to induce kindling. 193 Downloaded by guest on September 30, 2021 194 Neurobiology:. Valdes et al. Proc. Nad Acad. Sci.- USA 79 (1982) Table 1. [3H]Flunitrazepam binding in fascia dentata and immersed in 0.16 M Tris citrate, pH 7.1 at 4°C/2 nM hippocampal gyrus membranes after repeated seizures [3H]flunitrazepam; some solutions contained 2.5 ,um diazepam Fascia dentata Hippocampal gyrus to generate sections with nonspecific binding only. After in- cubation for 40 min at 0°C, the slide-mounted sections were Control vs. 551 ± 27 353 ± 6 in electroshock 735 ± 33* 336 ± 6 washed for 2 min buffer to reduce nonspecific binding. The (n = 7) (n = 13) slides were then rapidly dried under a stream of cold air, and Control vs. 408 ± 24 288 ± 15. flexible emulsion [Kodak NTB2 emulsion/water (1:1)]-coated kindled 576 35t -317 ± 20 coverslips were attached with glue. After exposure for 10-14 (n =21) (n = 19) days, the coverslips were gently bent away from the tissue sec- tion, the latent autoradiograms were developed, and the tissue Results are expressed in fmol of [3Hlflunitrazepam bound/mg of was fixed and stained with cresyl violet. The assemblies-were protein (mean ± SEM). Determinations in fascia dentata of the elec- and then examined troshock-treated group were obtained from the right hippocampal for- reapposed with both light- and dark-field mation of seven pairs of electroshock and control rats; hippocampal optics. gyri (n = 13) were obtained from both left and right hippocampal for- Preliminary experiments showed that binding to slide- mation of the seven pairs. Fascia dentata and hippocampal gyri were mounted sections had properties similar to-binding to mem- obtained from both hippocampal formations of 11 pairs of kindled and branes. Coronal sections (16 ,um) of rat cerebral hemispheres control rats. Data from the left and right sides of kindled animals were were used. After incubation, the portion ofthe slide containing pooled because no significant differences were found between the two. the tissue was snapped off and in Protosol to Analysis by Student's t test showed no significant differences between placed overnight seizure-treated and control animals except where specified. solubilize the tissue, and the radioactivity was determined in *P < 0.005. Omnifluor. Specific binding accounted for at least 95% of-total tP < 0.001. binding. The properties of specific binding to these slide- mounted sections were similar to those reported previously by [3H]flunitrazepam binding. in the absence and presence of 2.5 Young and Kuhar (12) and to binding with membranes with re- ,uM diazepam; it accounted for 80-95% of total binding under spect to kinetics, binding isotherms, and displacement with the conditions used. The incubation mixture (vol, 1.5 ml) was nonradioactive diazepam. 50mM Tris citrate buffer, pH 7.1/0.027% ethanol/2.0 nM [3H]- flunitrazepam containing 50-100 ,g of membrane protein. RESULTS Duplicate assays were conducted in the absence and presence Localization of Receptor Increases to Fascia Dentata: Mi- of2.5 ,uM diazepam. The reaction was continued for 80 min at crodissection Techniques. Repeated kindled seizures induced 0°C and terminated by vacuum filtration through a Whatman a marked increase in [3H]flunitrazepam binding in membranes GF/B filter. The filter was washed three times with 5-ml por- from fascia dentata but not hippocampal gyrus (Table 1). In kin- tions ofice-cold 50 mM Tris citrate buffer. The radioactivity on dled animals receiving a mean of 20.9 stimulations (18-23), the filter was measured in an Intertechnique scintillation [3H]flunitrazepam binding was increased by 41% in comparison counter as described (6). Because all binding assays could not with electrode-implanted unstimulated controls. Analysis of be performed on the same day, control and experimental ani- these data showed significant differences (P < 0.001, Student's mals were paired before killing unless specified otherwise. All t-test). The slight increase (10%) of [3H]flunitrazepam binding preparative and experimental procedures were carried out with in hippocampal gyrus membranes was not significant. scrupulous adherence to this paired protocol. Localization of benzodiazepine receptor increases after re- Autoradiography. The autoradiographic procedure was peated electroshock seizures was similar to that after kindled adapted from Young and Kuhar (11). Hippocampal formation seizures. In electroshock-treated animals (a total of 19 seizures), was removed by blunt dissection from control and experimental [3H]flunitrazepam binding was increased by 33% in comparison animals as described above. The samples were frozen by slow with paired unstimulated controls (Table 1). Analysis of these immersion in isopentane cooled (-70°C) in liquid nitrogen. data by Student's t test showed significant differences (P < Thin (6- to 12-,m) sections were cut in a cryostat microtome 0.005, two-tailed). By contrast, a slight nonsignificant decrease and thaw mounted onto gelatin-coated slides. Sections for study of [3H]flunitrazepam binding was found' in hippocampal gyrus were restricted to those removed from the middle one-third of membranes. the longitudinal axis ofhippocampal formation. The slides were To determine whether the increase of [3HH]flunitrazepam

Grains, no./625 gm2 0 10 20 30 40 50 60 70 80 90 100 ORIENS

PYR AM IDA LE

RADIATUM

LACUNOSUM MOLECULARE

MOLECULAR LAYER ii i g II II . .

GRANULE CELL - LAYER P OLYM ORPH E

FIG. 1. (Left) Cross section of rat hippocampal formation with subfields denoted according to Lorente-de-No (14). (Center) Laminae of area CA1 and fascia dentata according to Ramon y Cajal (15) shown in relationship to a schematic pyramidal (CA1) or granule (fascia dentata) cell. (Right) Silver grain counts ofbenzodiazepine receptor autoradiographs from control and experimental animals (see Fig. 2). Grain counts-in sections analyzed for nonspecific binding were <1/625 ,um mu, Experimental; _, control. Downloaded by guest on September 30, 2021 Neurobiology: Valdes et al. Proc. NatL Acad. Sci. USA 79 (1982) 195

binding was due to alteration in affinity or in number ofbinding previous observations on benzodiazepine binding isotherms in sites, the following experiment was carried out. Membranes membranes from whole hippocampal formation (6), these data were prepared from fascia dentata of electroshock-treated and demonstrate that the increases in [ H]flunitrazepam binding in control animals. Binding was measured at various concentra- fascia dentata in the paired protocol were due to increased num- tions of [3H]flunitrazepam and maximal values (Bm.) and k bers of binding sites. values were determined by linear regression analyses of Scat- Localization of Benzodiazepine Receptor Increases in Fas- chard plots (13). The results showed a 28% increase in the total cia Dentata: Radiohistochemical Techniques. The relatively number ofbinding sites in electroshock-treated compared with simplified anatomy, together with the lamination of the neu- control animals (952 ± 43 fmol/mg ofprotein versus 745 ± 57). ronal constituents, in fascia dentata offers important advantages This difference was statistically significant (P < 0.0004, Stu- for localizing a molecular alteration to a cellular element. The dent's t test). The Kd values for the electroshock-treated (0.82 somata of the principal neuron, the granule cell, are arranged ± 0.08 nM) and control (1.04 ± 0.16 nM) groups were not sig- in a tightly packed layer termed the granule cell layer. The den- nificantly different. Similar results were obtained in an exper- drites extend perpendicularly from the plane of the cell body iment comparing kindled and control groups. Together with our layer and ramify throughout the entire thickness ofthe molec- ular layer (Fig. 1). The major excitatory afferents arise from LM M G H G M ; afferents from lateral and medial entorhinal cortex form synapses on granule cell dendrites in the outer and middle thirds of the molecular layer, respectively. The quan- , titatively less significant commissural and associational afferents form synapses on granule cell dendrites in the inner third ofthe .. molecular layer. If the seizure-induced increases of benzodi-

's azepine receptors in fascia dentata are localized to a single pop- ulation of neurons, the laminar pattern on autoradiographs er ~~~~J- 1 ;Qt could provide a clue to the identity of the involved neurons. To determine the localization of the seizure-induced in- ' . creases of benzodiazepine receptors within fascia dentata, .N AAM [3H]flunitrazepam autoradiographs were developed from thin Vfrk (12-,um) sections from hippocampal formation of electroshock- treated and control animals. These autoradiographs demon- strated increases in the number ofsilver grains over the granule cell body and throughout the entire molecular layer (Fig. 2). The increased numbers of silver grains were apparent on low- power magnification photomicrographs (dark field) of sections from control and seizure-treated animals. These increases were verified and quantitated by grain counts through a calibrated eyepiece underhigh-power (x 400) magnification (Table 2; Fig. 1). No marked differences were noted in dentate hilus, stratum lacunosum moleculare of CAl, or any other subregion of hip- pocampal formation. Similar results were obtained with mul- tiple sections prepared from both hippocampal formations of one pair ofa control and an electroshock-treated animal and two pairs of a control and a kindled seizure-treated animal. The striking distribution of the increased grain density throughout the molecular and granule cell layers is most consistent with localization to the somata and dendrites of the granule cells.

Table 2. Localization of benzodiazepine receptor increases after repeated seizure treatments: Silver grain counts from light microscopic autoradiographs Region Control Electroshock % control Hilus 26.4 ± 1.2 30.1 ± 1.5 114 Granule cell layer 27.0 ± 1.8 44.5 ± 0.8* 165 Molecular layer 53.7 ± 1.9 84.2 ± 2.1* 157 Stratum lacunosum moleculare 47.1 ± 2.0 49.0 ± 1.5 104 Stratum radiatum 49.7 ± 1.9 53.1 ± 2.0 106 Stratum pyramidale 40.1 ± 2.0 37.6 ± 1.8 94 Stratum oriens 54.9 ± 1.9 51.8 ± 2.1 94 FIG. 2. (Upper) Light-field photomicrograph of cresyl violet stain Results are expressed as number of silver grains per 625 ,um2 in cross section of fascia dentata as pictured in dark field. (x43.) Dark- counted through an eyepiece fitted with a reticle and are mean ± SEM field photomicrographs of [3Hlflunitrazepam autoradiographs from for 15 determinations. The slides analyzed and the photomicrographs control (Middle) and seizure-treated (Lower) fascia dentata. Note the (Fig. 2) were obtained from the same animal. The number of silver increased density of silver grains over granular cell andmolecular lay- grains in sections incubated with both [3Hlflunitrazepam and nonra- ers of experimental relative to control. Regions: LM, stratum lacu- dioactive diazepam was 0.7 ± .1 and did not differ between experi- nosum moleculare of CA1; M, molecular layer of fascia dentata; G, mental and controls. granule cell layer of fascia dentata; H, dentate hilus. * Significant (P < 0.001, Student's t test, two-tailed). Downloaded by guest on September 30, 2021 196 Neurobiology: Valdes et al. Proc. Nad Acad. Sci. USA 79 (1982)

Table 3. Effects of colchicine-mediated granule cell destruction on fascia dentata benzodiazepine receptors Biochemical analyses (n = 6) Silver grains, no. per [3H]Flunitrazepam [3H]Flunitrazepam 625 Jm2 (n = 15) binding, fmol/mg Membrane protein, pg binding to fascia Granule Molecular of-protein per fascia dentata dentata, fmol cell layer layer Control 578 ± 27 783 ± 45 452 38 ± 2 57 ± 2 Colchicine 289 ± 48* 237 ± 34* 73 18 ± 1* 25 ± 1* % control 47 44 Results are mean ± SEM. [3H]Flunitrazepam binding to fascia dentata was calculated by multiplying [3Hlflunitrazepam binding (fmol/mg of protein) by protein content per fascia dentata. * Significant (P < 0.001, Student's t test, two-tailed).

Effects ofGranule Cell Destruction on Benzodiazepine Re- Effects of Granule Cell Destruction by Intrahippocampal ceptors in Fascia Dentata: Microdissection Techniques. Lo- Colchicine on Benzodiazepine Receptors: Radiohistochemical calization ofbenzodiazepine receptor increases to granule cells Techniques. Ifelimination ofgranule cells is the mechanism by suggests that granule cells normally contain benzodiazepine which colchicine produces a loss of fascia dentata benzodiaze- receptors. To test this suggestion, granule cells were destroyed pine receptors, the distribution of silver grain loss should cor- by two independent methods and benzodiazepine receptor respond to the cell bodies and dendrites ofgranule cells in fascia bindingwas measured in membranes from microdissectedfascia dentata. Such a distribution was found. Significant reductions dentata. of silver grains were found over both the granule cell and mo- Previous granule cell destruction by intrahippocampal injec- lecular layers (Table 3). This further supports the idea that gran- tion of colchicine caused a marked loss of benzodiazepine re- ule cells contain a considerable portion of benzodiazepine re- ceptor binding sites in fascia dentata membranes. To destroy ceptors in fascia dentata. granule cells, colchicine was injected (6 nmol per injection site) The majority (86%) ofsynapses in the outer two-thirds ofthe bilaterally into both septal and ventral hippocampal formation molecular layer represent entorhinal cortical afferents termi- as described (16). Animals were sacrificed ==1 week later. Con- nating on granule cell dendrites (18). Axons ofthese entorhinal trol animals were anesthetized but received no injections. The cortical neurons course through the angular bundle en route to resulting granule cell destruction was verified histologically their terminations. To test whether the presynaptic terminals (data not shown) in frozen sections cut from slices removed from ofthese afferents contained benzodiazepine receptors, the fol- each hippocampal formation. Paralleling this marked granule lowing experiment was performed. The right angular bundle cell destruction was a 50% reduction of[3H]flunitrazepam bind- was cut with a knife with stereotaxic guidance under pentobar- ing (Table 3). This decrease was significant (P < 0.001, Student's bital anesthesia. In control animals, the skull was trephined but t test). Binding isotherms in additional membranes showed this no knife cut was made. The animals were sacrificed 2 days later, decrease to be due to decreased numbers ofbinding sites with- and the brains were serially sectioned. Evidence ofdenervation out significant alterations in affinity (Bm., control 1320 ± 123 (19) in the dorsal right hippocampal formation was confined to versus colchicine 525 ± 34 fmoVmg;Kd, control 1.0 ± 0.2 ver- the middle third ofthe molecular layer in the experimental an- sus colchine 0.9 ± 0.1 nM). This reduced binding was not sim- imal, which corresponded to the cutting ofafferents from medial ply accounted for by the presence of colchicine; inclusion of but not lateral entorhinal cortex (20). Examination of [3H]- colchicine (4 ,uM) in the incubation mixture caused no consist- flunitrazepam autoradiographs ofadjacent sections disclosed no ent alterations ofKd or Bm. value of[3H]flunitrazepam binding decrease in silver grain density over the middle third of the to brain membranes. The 50% reduction in number of molecular layer in comparison with either the inner or the outer [3H]flunitrazepam binding sites in the presence of only 30% third (not denervated) of the molecular layer on the same side remaining protein indicates a loss of 85% of [3H]flunitrazepam as the lesion nor in comparison with the side without lesions or binding sites per fascia dentata in the colchicine-treated group. control animals. Moreover, no "piling up" of silver grains was To determine whether granule cell destruction caused by a found in the area of the knife cut, as might be expected if re- different method was associated with reduced [3H]flunitrazepam ceptors were undergoing axoplasmic transport from the cell binding, suckling rats were irradiated [200 rads (1 rad = 0.01 bodies (21). These data argue against the presence of benzo- gray) on postnatal days 2 and 3, 150 rads on days 5, 7, 9, and diazepine receptors in presynaptic terminals and thereby 11] (17). Littermates were handled but not irradiated. The an- strengthen the localization to granule cells. imals were sacrificed ""2 months later. The irradiation effects were histologically verified in frozen sections from hippocampal formation slices removed at the time ofdissection. This showed DISCUSSION significant loss of granule cells (data not shown) and The principal findings of this study pertain to the localization [3H]flunitrazepam binding [control 450 ± 19 (n = 12) versus of the increased numbers of hippocampal formation benzodi- irradiated 314 ± 21 (n = 9), P < 0.005]. The decrease in protein azepine receptors induced by repeated seizures. Biochemical content in the fascia dentata was used as an index ofgranule cell analyses of microdissected membranes localize the increase loss. The data for both colchicine and radiation treatment were mainly to fascia dentata. Assuming the increased receptors re- plotted as [fmol bound (treated)/fmol bound (control)] versus side in a single cell population within fascia dentata, the re- [mg of protein per fascia dentata (treated)/mg of protein (con- markable anatomic specificity found radiohistochemically ar- trol)]. This generated a straight line of slope 0.9 and protein gues for localization to the somata and dendritic tree of the intercept 0.13 (P < 0.001), which supports a 1:1 relationship granule cells. This idea is supported by lesion experiments in- between granule cells and benzodiazepine receptors. The pos- dicating that granule cells normally contain benzodiazepine re- itive intercept suggests that some elements offascia dentata do ceptors. Moreover, the presynaptic terminals ofthe entorhinal not contain benzodiazepine receptors. afferents appear to be devoid ofbenzodiazepine receptors. Lo- Downloaded by guest on September 30, 2021 Neurobiology: Valdes et al. Proc. Natl. Acad. Sci. USA 79 (1982) 197

calization to other cellular constituents is unlikely, either be- dling (i.e., fewer stimulations required to induce subsequent cause of the relative paucity of cells with overlapping distri- kindling compared with non-seizure-treated controls) (29, 30). bution (e.g., basket cells) or because the normal distribution Understanding the precise role ofthe benzodiazepine receptor does not conform to this discrete pattern (e.g., glia, blood increases in kindled epileptogenesis must await elucidation of vessels). the endogenous ligand for this receptor and its physiologic The cellular and molecular mechanisms underlying the kin- actions. dling phenomenon are obscure (7). A critical limiting factor is The present findings correlate biochemical and morphologic our ignorance of the spatial distribution of the altered neural results pinpointing a molecular alteration to a specific neuronal circuitry responsible for kindling. Evidence suggests that kin- population in an animal model of epilepsy. This is an important dling is subserved by alteration of a network of neural circuits step in determining how this molecular alteration translates into that has a highly specific distribution, both in the area of the altered neuronal function. The findings point to the dentate stimulating electrode and in "downstream" synaptic sites (9). granule cells as a site for biochemical, electrophysiologic, and A proposed explanation for the facilitation and amplification of the basic amyg- information flow through these circuits is the development of morphologic investigations into mechanisms of long-term potentiation of synaptic communication (22, 23). dala kindling. The presumptive localization of benzodiazepine receptor in- We thank Ms. Eloise Pittman for her secretarial assistance. This work creases to the granule cells is particularly interesting within this was supported by Veterans Administration Research Grant 392-42- framework for two reasons. First, the responsiveness ofgranule 0819-2 and National Institutes of Health Grants NS-16431 and NS- cells to their principal excitatory inputs readily undergoes long- 06552. term potentiation (23). Second, the granule cell is a critical relay in information flow through hippocampal formation. Cortical 1. Racine, R., Livingston, K. & Joaquin, A. (1976) Electroen- and many subcortical regions, including amygdala, gain access cephalogr. Clin. Neurophysiot 38, 355-365. with 2. Mattson, R. H. (1972) in Antiepileptic Drugs, eds. Woodbury, D. to hippocampal formation circuitry through connections M., Penry, J. K. & Schmidt, R. P. (Raven, New York), pp. entorhinal cortex, which in turn provides an excitatory synaptic 497-516. input to the granule cells (24, 25). The granule cell represents 3. Squires, R. F. & Braestrup, C. (1977) Nature (London) 266, the first link of a feed-forward excitatory chain in hippocampal 732-734. formation sequentially activating CA3 and CAI pyramids (25). 4. Goodard, G. V., McIntyre, D. C. & Leech, C. K. (1969) Exp. In vivo electrophysiologic studies indicate that granule cells are Neurol 25, 295-330. 5. Wada, J. A. (1976) Kindling (Raven, New York). a key site in regulating information flow through this trisynaptic 6. McNamara, J. O., Peper, A. M. & Patrone, V. (1980) Proc. Natl chain under physiologic conditions (26). Acad. Sci. USA 77, 3029-3032. The granule cells may also regulate information flow through 7. McNamara, J. O., Byrne, M. C., Dasheiff, R. M. & Fitz, J. G. these circuits under pathologic conditions such as seizures. Prog. Neurobiol. 15, 139-159. Exposure of hippocampal slices to convulsant agents in vitro 8. Racine, R. J. (1972) Electroencephalogr. Clin. Neurophysiol 32, results in burst firing ofCA3 and CAL pyramids, a firing pattern 281-294. 9. Messenheimer, J. A., Harris, E. W. & Steward, 0. (1979) Exp. observed in neurons in epileptic foci (27). By contrast, granule Neurol 64, 469-481. cells themselves do not burst under these conditions. When 10. Lowry, 0. H., Rosebrough, N. J., Farr, A. L. & Randall, R. J. slices are exposed to normal media, bursting of CA3 pyramids (1951)j. Biol. Chem. 193, 265-275. can be induced by excitatory synaptic input from granule cells 11. Young, W. S. & Kuhar, M. J. (1980) Brain Res. 179, 255-270. (27). It seems likely that enhanced synaptic drive of CA3 pyra- 12. Young, W. S. & Kuhar, M. J. (1980)J. Pharmacol Exp. Ther. 212, 337-346. mids, induced by long-term potentiation of excitatory synaptic 13. Scatchard, G. (1949) Ann. N.Y. Acad. Sci. 51, 660-672. activation ofgranule cells, could play a critical role in triggering 14. Lorente-de-No, R. J. (1934) Psychol Neurol 46, 113-117. limbic seizures. Likewise, a molecular alteration designed to 15. Ramon y Cajal, S. (1893) AnaL Soc. Esp. Hist. Nat. (Madrid) 22, reduce granule cell excitability could protect against limbic 53-114. 16. Goldschmidt, R. B. & Steward, 0. (1980) Proc. Nati. Acad. Sci. seizures. USA 77, 3047-3051. Whether increased benzodiazepine receptors result in in- 17. Altman, J. & Anderson, W. J. (1971) Exp. Neurol 30, 492-509. creased or decreased granule cell excitability is unclear. The 18. Matthews, D. A., Cotman, C. W. & Lynch, G. (1976) Brain Res. increased numbers of benzodiazepine receptors could repre- 115, 1-21. sent any of three major processes in kindled rats: (i) the molec- 19. Fink, P. R. & Heimer, L. (1967) Brain Res. 4, 369-374. 20. Hjorth-Simonsen, A. (1972)J. Comp. Neurol 146, 219-232. ular basis of kindling, (ii) the molecular basis of a process re- 21. Laduron, P. (1980) Nature (London) 286, 287-289. sisting development of kindled epilepsy, or (iii) a molecular 22. Racine, R. J., Gartner, J. G. & Burnham, W. M. (1972) Brain consequence ofrepeated seizures with no meaningful functional Res. 47, 262-268. significance regarding the likelihood of subsequent seizures. 23. Douglas, R. M. & Goddard, G. V. (1975) Brain Res. 86, 205-215. 24. Krettek, J. E. & Price, J. L. (1977) J. Comp. Neurol 172, The anatomic specificity of the receptor localization argues 723-752. against the third possibility. Arguing in favor of one of the first 25. Anderson, P., Holmavist, B. & Voorhoem, P. E. (1966) Acta two possibilities is our observation that previous granule cell Physiol. Scand. 66, 461-472. destruction by intrahippocampal colchicine retards the subse- 26. Winson, J. & Abzug, C. (1978) J. Neurophysiol. 44, 937-950. quent development of amygdala kindling (28). The occurrence 27. Prince, D. (1978) Annu. Rev. Neurosci. 1, 395-416. 28. Dasheiff, R. M. & McNamara, J. 0. (1981) Ann. Neurol., in press. of the receptor increases after repeated electroshock seizures 29. Kilbey, M. M., Ellinwood, E. M. & Easler, M. E. (1979) Exp. does not aid in differentiating the first two possibilities, because Neurol. 64, 306-314. repeated seizures induced by other agents result in partial kin- 30. Cain, D. P. (1981) Neurosci. Abstr. 7, 586. Downloaded by guest on September 30, 2021